NASA Sets Launch Pad for HPC, ML Startups

There are few better ways to land on the radar than having an ongoing contract with NASA for systems and software.

The space agency has selected 289 startups and small companies as part of its Small Business Innovation Research (SBIR) Program in its first phase, highlighting the work of several startups and established small companies that bridge the high performance computing and machine learning divide. NASA selects the companies based on technical merit and commercial potential.

While certainly not all of the selected companies play in the HPC/AI space, companies like MemComputing, which NASA selected for its virtual MemComputing Machine (VMM) technology highlight the agency’s focus on exploring non-standard data and system architectures. The startup has a varied SaaS-based platform via VMM that is based on the company’s novel circuit architecture. They say they can solve physics problem with quantum computing speed and that for NASA, they can snap into modeling and simulation areas handled on traditional supercomputers and solve problems more cost-effectively and efficiently.

EP Analytics has developed a general purpose and cross-platform way to address deployment of NASA’s rich base of HPC codes. According to the company the platform will streamline the deployment of complex modeling and simulation (M&S) applications on a diverse set of computing resources (local desktop/laptops, all major cloud-service providers and traditional on-premise HPC clusters). SimplifyHPC, is a desktop application that brings together all the tools necessary for M&S scientist and will consist of “a built-in performance monitoring ensuring optimal deployment of applications.”

Tactical Computing Laboratories is another NASA pick for RISC-V based technologies. The company says it will be addressing the needs of image processing and other data parallel scientific applications with its RadRISC platform, a scalable multicore RISC-V based architecture. They will deploy an integrated development environment for RadRISC for NASA using their open source SoC development infrastructure. They say the initial system architecture will include an arrangement of RISC-V cores connected via RapidIO fault tolerant switching connected to memory. While this is for more autonomous and robotic technologies, Tactical Computing Labs is known in HPC and RISC-V advocacy circles for its potential in larger scalable systems.

Continuum Dynamics is another small company familiar to some in the HPC space, especially those in government circles. They focus on specific scientific computing and engineering problems in aerospace and defense and while not necessarily a startup (the company was founded in 1979), they tend to fly under the radar. For the NASA work they’ll focus on code modernization for a mission-critical piece of software core to NASA’s heliophysics modeling with an emphasis on code performance, portability, and usability.

Houston-based Tietronix Software will be developing a library of reusable NASA SysML models. The company is focused on software engineering services for government and defense applications. The co-founder, Victor Tang, is familiar with NASA after spending years working via IBM and Wang Government Services to build software for the space agency. The other co-founder was NASA’s Chief of Space Flight Training with 33 years logged at NASA. The company is planning to use a blockchain technology approach to enhance model-based engineering at NASA for better security and transparency of their distributed environment.

The agency provided up to $125,000 for companies to establish the merit and feasibility of their innovations. Phase I SBIR contracts are awarded to small businesses and last for six months, while Phase I STTR contracts are awarded to small businesses in partnership with a research institution and last for 13 months. Based on their progress during Phase I, companies may submit proposals to subsequent SBIR/STTR opportunities and receive additional funding.